Nvidia and 36 Friends Form Open Secure AI Alliance

Other members of the group include Microsoft, Cisco, Cloudflare, Hugging Face, IBM and Red Hat. The idea is to develop and share open technologies, techniques and tools for securing software and artificial intelligence agents.

Ponder this. Why would companies that exist only to sell you software give away this valuable AI tools. That doesn’t mean the results won’t be appreciated. It is just that the motive is not to be altruistic.

They plan to cover the full agent stack and secure coding workflows.

The spin is that defenders need AI models that they can change and run locally.

After all, I am sure that everyone is ready to share all of their security event data with say (just picking on one guilty party) OpenAI. Not likely.

The first contribution is Nvidia’s OO Agents (NOOA), an Apache Research framework designed to make agent behavior easier to test, trace, audit and govern.

Even when this is actually released it probably won’t be useful for small or medium businesses that don’t have sophisticated IT teams, but hopefully, some of the other deliverables will be useful.

We know that the bad actors are already using AI; it would be helpful if the defenders had the same capabilities. Credit: The Hacker News

by